Description:

There is a troubling divergence between many of our intuitions for the benefits of learning with perceptually grounded, highly contextualized and concrete pedagogical materials and empirical results suggesting benefits for idealized, decontextualized, and formalized materials, particularly for transfer of learning to new situations. In an effort to reconcile intuitions and empirical results, a delineation can be made between several ways in which knowledge can be grounded in perception and action, and the opportunities of each for facilitating transfer. Consistent with this delineation, some positive cases of perceptually grounded yet transferable knowledge in math and science will be described. Finally, as a prelude to predicting transfer successes and failures, I will describe a computational model (SUBLIME) of the perception-action processes by which human learners discover patterns in a natural phenomenon by interacting with it or simulations of it. SUBLIME can serve as an automatic method for discovering not only simple patterns in nature, but likely misconceptions that students will generate as they look for these patterns.